Assessing protection forest structure with airborne laser scanning in steep mountainous terrain

Protection forest management requires reliable data on the structural characteristics of forest stands with high spatial resolution, which could be delivered by airborne laser scanning. We subtracted a digi-tal surface model (DSM), derived from the last LiDAR pulses, from a digital terrain model (DTM), de-rived from the first LiDAR pulses, to obtain a normalized crown model (nCM). The resolution of the rasters was 1 m × 1 m. With two methods that are based on local maxima identifiers individual tree tops were detected with a mean error of 33% when comparing the number of detected trees with the measured number of trees present in the validation plot. When only taking into account the dominant and co-dominant trees, this error decreased to approx. 10%. Position errors of the trees that were automatically identified in the nCM were between 0.5 and 3.5 m, when comparing to on-site GPS measured positions. Field investigations on the causes of errors in the number of detected trees showed that they are mainly caused by trees growing in collectives. Errors in tree positions are related to tilted trees and missed' tree tops during scanning, as well as the cumulative errors between GPS measured positions of the base of trees and the LiDAR position. In conclusion, we are of the opinion that airborne laser scanning provides excellent data for protection forest management. It provides reli-able information on the positions of individual, dominating and co-dominating trees and on the position of collectives and the tree heights. In addition, it provides excellent input data for 3D natural hazard simulation models, even in steep terrain.